Method and apparatus for the compression and/or transport and/or decompression of a digital signal

Abstract

A method is described for compressing and/or transporting and/or decompressing a digital signal. The most significant bits of a sample of the digital signal are compressed and transported as a compressed transport sample. Methods are used to improve the compression and transport as hybrid DPCM and a dynamic shift of clip range of prediction errors. For every sample, not only a compressed transport sample but also a residual transport sample is transmitted. The residual transport sample is either equal to the least significant bits of the sample or equal to a substitution value which is a function of the clipping error from the compression of the most significant bits of the said sample. Apparatuses are described for compression and/or transport and/or decompression of a digitized television IF signal according to the method.

1. Method for the transmission of a first digital signal (DS 1 ) from a first location (FL) over a transport channel (TC) to one or more second locations (SL) where it is received as a second digital signal (DS 2 ) which is substantially equal to the first digital signal (DS 1 ), whereby a sample (S 1 ) of the first digital signal (DS 1 ) represented by M bits, being the total of N most significant bits (MSB) and M-N least significant bits (LSB), is transported over the transport channel (TC) as a transport sample comprising at least two parts, of which
 one part is a compressed transport sample (CTS) represented by N-C bits, N being smaller than M and C being smaller than N and larger than 0, the N-C bits being obtained through predictive coding of the N most significant bits of the sample (S 1 ) of the first digital signal (DS 1 ), whereby to each sample (S 1 ) of the first digital signal (DS 1 ) corresponds at least a prediction ({circumflex over (x)} enc ) representing the predicted N most significant bits of the sample (S 1 ) of the first digital signal (DS 1 ) whereby the prediction is based on previously compressed samples, a prediction error (e enc ) representing the difference between the N most significant bits of the sample (S 1 ) of the first digital signal (DS 1 ) and said prediction ({circumflex over (x)} enc ), and a clipping error representing the difference between the prediction error (e enc ) and a clipped prediction error ((e enc ) c ) which is the prediction error (e enc ) clipped by a first clipper (PEC 1 ) to a clip range represented as [A . . B] by means of N-C bits, A and B being integers and B-A being equal to or smaller than 2 (N-C) −1, and  
 the other part is a residual transport sample (RTS) represented by M-N bits, characterized in that said residual transport sample (RTS) is either equal to the M-N least significant bits (LSB) of said sample (S 1 ) of the first digital signal (DS 1 ) in the case that the prediction error (e enc ) corresponding to said sample (S 1 ) of the first digital signal (DS 1 ) is in the range [A . . B], or, in the other case, the P least significant bits of the M-N LSB's are equal to a substitution value (CE) which is a quantization function of the absolute clipping error corresponding to the said sample (S 1 ) of first digital signal (DS 1 ), whereby the number of output levels of said quantization function of the absolute clipping error is equal to or less than 2 P , whereby P is equal to or less than M-N and whereby the other M-N-P bits of the M-N bits of the RTS are equal to the M-N-P most significant bits of the M-N LSB's of said sample (S 1 ) of the first digital signal (DS 1 ).  
 
   
   
     2. Method according to  claim 1 , wherein the compressed transport sample (CTS) is the clipped prediction error ((e enc ) c ). 
   
   
     3. Method according to  claim 1 , wherein the compressed transport sample (CTS) is an in the range [A . . B] wrapped around sum of the clipped prediction error ((e enc ) c ) and a mapped value (m({circumflex over (x)} enc )) of the prediction ({circumflex over (x)} enc ), and that said mapped value (m({circumflex over (x)} enc )) of the prediction ({circumflex over (x)} enc ) is the prediction ({circumflex over (x)} enc ) mapped on a range [D . . E], E and D being integers and E-D being equal to or smaller than 2 (N-C) −1. 
   
   
     4. Method according to  claim 3 , wherein the prediction ({circumflex over (x)} enc ) is mapped or quantized in a non-uniform way, such that the quantization is fine for prediction values corresponding to small input amplitudes and rough for prediction values corresponding to big amplitudes of the first digital signal. 
   
   
     5. Method according to  claim 1 , whereby the first digital signal (DS 1 ) is a digitized television IF signal with an IF carrier and featuring as parameters at least an estimated luminance and an estimated IF carrier phase, the second digital signal (DS 2 ) consequently also being a digitized television IF signal, wherein:
 the clip range [A . . B] is shifted over a shift (sh) to [A+sh . . B+sh], whereby the shift (sh) for a sample is determined by the estimated luminance and/or by the estimated IF carrier phase in that sample,  
 the absolute amount of the shift (sh) is proportional to the absolute value of the amplitude of the estimated IF carrier,  
 for a television IF signal with negative modulation, either the shift (sh) is negative in case of a positive peak of the estimated IF carrier and a low luminance, and in case of a negative peak of the estimated IF carrier and a high luminance, or the shift (sh) is positive in case of a negative peak of the estimated IF carrier and a low luminance, and in case of a positive peak of the estimated IF carrier and a high luminance,  
 for a television IF signal with positive modulation, either the shift (sh) is positive in case of a positive peak of the estimated IF carrier and a low luminance, and in case of a negative peak of the estimated IF carrier and a high luminance, or the shift (sh) is negative in case of a negative peak of the estimated IF carrier and a low luminance, and in case of a positive peak of the estimated IF carrier and a high luminance.  
 
   
   
     6. Method according to  claim 5 , wherein the compressed transport sample (CTS) is in the range [A . . B] wrapped around sum of the clipped prediction error ((e enc ) c ) and a mapped value m(y) with y being either the prediction {circumflex over (x)} enc  or the sum of the prediction {circumflex over (x)} enc  and the shift sh, and that said mapped value m(y) is in a range [D . . E], E and D being integers and E-D being equal to or smaller than 2 (N-C) −1, the mapping being either a uniform or a non-uniform mapping, the non-uniform mapping being such that the quantization is fine for prediction values corresponding to small input amplitudes and rough for prediction values corresponding to big amplitudes of the first digital signal.